Resource Monitoring

[DavidPL1]

The option to request resource information about components and hosts was requested. This could be achieved by writing a new module fetching data and keeping stats that is called by the monitoring thread to update periodically.

Options to fetch the data are:

• psutil (already dependency): able to get process cpu and memory stats
• watchdog: directory monitoring (could monitor /proc/pid for changes) but as own thread
• bash scripting: extract information from top or with ps

However those options only work for local monitoring. For remote monitoring the hosts have to communicate with each other (remote host runs own mon thread and reports to master or master invokes monitoring and fetches results by itself). Using only ssh may not be the best choice since parsing cli output feels unsafe and dirty. A better way could be a dedicated connection for this purpose over a tcp or udp socket or letting a library take care of the low-level part of the connection and only use high-level features for communication (e.g. twisted). This could also be used as a safer way of performing remote component and host availability checks.

[DavidPL1]

Forgot to ref this in 5de1e969eadc8fb2c329163b0f8baeaabccd71e7, 5c6f6e5321c39b8fe628237b8d982c55288e652e, 5d7704878cbe3ca6d4780bb7fe1d6194b255ea13 and ad9703b66cb0a5c273651eae4bf8c71f87cfdb86.

5de1e969eadc8fb2c329163b0f8baeaabccd71e7 pretty much resolves this.